Information processing device, information processing method and storage medium

Abstract

The present invention relates to an information processing device, an information processing method and a storage medium. The information processing device acquires a material structure image representing the structure of a material, and the material structure image is obtained by imaging the material. The information processing device generates a Fourier transform result representing a power spectrum by applying Fourier transform to the acquired material structure image. The information processing device generates a size distribution of a structure forming the material using an expectation maximization algorithm based on the Fourier transform result of the material structure image.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application is based upon and claims the benefit of priority under 35 USC 119 to Japanese Patent Application No. 2020-119514, filed on Jul. 10, 2020, the disclosure of which is incorporated herein by reference. Technical Field

[0003] The present disclosure relates to an information processing device, an information processing method, and a storage medium storing an information processing program. Background Art

[0004] To date, microscopic examination processes for grain size in steel are known. For example, the process disclosed in Japanese Industrial Standard JIS G0551:2013 "Steel - Microscopic determination of apparent grain size" measures the size of grains on the polished surface of a test piece processed by appropriate processing with a microscope, based on the type of steel and other information.

[0005] When the structural image of the material obtained by the microscope disclosed in the above-mentioned reference is used to detect the size of the structure of the material shown in the image, a lot of time and labor are required. Therefore, this conventional technology may not be able to effectively obtain the size distribution of the structure of the material. Summary of the invention

[0006] In view of the above circumstances, the present disclosure provides an information processing apparatus, an information processing method, and a storage medium storing an information processing program, which can effectively obtain the size distribution of structures forming a material from a configuration image of the material.

[0007] According to the first aspect, the information processing device includes: an acquisition unit, which acquires a material structure image representing the formation of a material, wherein the material structure image is obtained by imaging the material; and a distribution generation unit, which generates a Fourier transform result representing a power spectrum by applying Fourier transform to the material structure image obtained by the acquisition unit, and based on the Fourier transform result of the material structure image, uses an expectation maximization algorithm to generate a size distribution of a structure forming the material.

[0008] The information processing device acquires a material structure image representing the structure of the material, which is obtained by imaging the material. Then, by applying Fourier transform to the material structure image, the information processing device generates a Fourier transform result representing a power spectrum. Based on the Fourier transform result of the material structure image, the information processing device uses an expectation maximization (EM) algorithm to generate a size distribution of structures forming the material. Therefore, the size distribution of structures forming the material can be effectively obtained from the structure image of the material.

[0009] The information processing device according to the second aspect further includes an image generating unit, which generates a contrast image including multiple material structure images with different contrasts based on the material structure image acquired by the acquisition unit, wherein the distribution generating unit generates a Fourier transform result for each of the multiple contrast images generated by the image generating unit, and based on the corresponding Fourier transform results of the multiple contrast images, an expectation maximization algorithm is used to generate the size distribution of the structure forming the material.

[0010] In the information processing method according to the third aspect, a computer performs processing, which includes: acquiring a material structure image representing the structure of the material, the material structure image being obtained by imaging the material; generating a Fourier transform result representing a power spectrum by applying Fourier transform to the acquired material structure image; and based on the Fourier transform result of the material structure image, using an expectation maximization algorithm to generate a size distribution of the structure forming the material.

[0011] According to the fourth aspect, a storage medium storing an information processing program stores an information processing program, which enables a computer to perform processing, the processing including: acquiring a material structure image representing the structure of the material, the material structure image being obtained by imaging the material; generating a Fourier transform result representing a power spectrum by applying Fourier transform to the acquired material structure image; and based on the Fourier transform result of the material structure image, using an expectation maximization algorithm to generate a size distribution of a structure forming the material.

[0012] According to the present disclosure described above, there is provided an effect in that the size distribution of structures forming a material can be efficiently obtained from a configuration image of the material. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is a schematic block diagram of an information processing apparatus according to a first exemplary embodiment.

[0014] Figure 2 is a diagram describing a conceptual diagram of the first exemplary embodiment.

[0015] Figure 3 is a diagram describing a conceptual diagram of the first exemplary embodiment.

[0016] Figure 4 is a graph depicting the estimation of size distribution based on Fourier transform results.

[0017] Figure 5 is a graph depicting the estimation of size distribution by the EM algorithm.

[0018] Figure 6is a graph depicting the estimation of size distribution by the EM algorithm.

[0019] Figure 7 is a graph depicting the estimation of size distribution by the EM algorithm.

[0020] Figure 8 is a diagram showing a configuration example of a computer of the information processing apparatus according to the exemplary embodiment.

[0021] Fig. 9 is a flowchart illustrating an example of processing performed at the information processing apparatus according to the first exemplary embodiment.

[0022] Fig.10 is a schematic block diagram of an information processing apparatus according to a second exemplary embodiment.

[0023] Fig.11 is a flowchart illustrating an example of processing performed at the information processing apparatus according to the second exemplary embodiment. DETAILED DESCRIPTION

[0024] First Exemplary Embodiment

[0025] Next, an information processing apparatus according to a first exemplary embodiment is described using the drawings.

[0026] Figure 1 1 is a block diagram showing an example of the functional structure of the information processing apparatus 10 according to the first exemplary embodiment. Figure 1 As shown, the information processing device 10 is provided with an acquisition unit 100 , an image generation unit 102 , and a distribution generation unit 104 .

[0027] The acquisition section 100 acquires a material configuration image, which is an image captured when a material is observed using a microscope. The material configuration image is an image depicting the configuration of a material, and is a grayscale image.

[0028] Based on the material configuration image acquired by the acquisition section 100 , the image generation section 102 generates a comparative image that is a plurality of material configuration images having different contrasts.

[0029] Figure 2 and Figure 3 1 is a diagram for describing a schematic diagram of the information processing apparatus 10 according to the first exemplary embodiment. Figure 2 As shown, from the material configuration image of the material M obtained by the microscope, the information processing apparatus 10 according to the first exemplary embodiment generates a contrast image I colored according to different contrasts. m,A ,I m,B and I m,C. Structures of different sizes appear in multiple contrast images I m,A ,I m,B and I m,C In each contrast image in FIG. 1 , the material is formed of a plurality of structures. For example, the structure is a grain, a needle, a fiber, a hole, etc., which form the material.

[0030] Therefore, the information processing apparatus 10 according to the first exemplary embodiment estimates that the m,A ,I m,B and I m,C The size of the structure in each contrast image in , and generate Figure 2 The size distribution S shown A , S B and S C . Size distribution S A From the contrast image I m,A Generated, size distribution S B From the contrast image I m,B Generated, and the size distribution S C From the contrast image I m,C These size distributions may be calculated both for the size in the horizontal direction and for the size in the vertical direction of the structures appearing in the contrast image.

[0031] Figure 3 1 is a diagram for describing in more detail a schematic diagram of the information processing apparatus 10 according to the first exemplary embodiment. Figure 3 As shown, when a material is constructed into an image ( Figure 3 When the pixel value distribution of the top image in the figure is divided into different contrasts, the generated Figure 3 The multiple comparison images I shown m,A ,I m,B and I m,C . Comparison image I m,A is the contrast image corresponding to region A in the pixel value distribution, and contrast image I m,B is a contrast image corresponding to region B in the pixel value distribution, and contrast image I m,C is a comparison image corresponding to region C in the pixel value distribution. The information processing device 10 generates a comparison image corresponding to each of the plurality of comparison images I m,A ,I m,B and I m,C The corresponding size distribution H A , H B and H C In the size distribution H A , H B and H C Middle, Peak M A1 、M A2 、MB1 、M B2 、M C1 and M C2 is the peak of the size distribution, which can be calculated by previously known techniques.

[0032] For each of the plurality of comparison images generated by the image generation unit 102, the distribution generation unit 104 estimates the size of the structure forming the material appearing in the comparison image. Based on the estimated size of the structure, the distribution generation unit 104 generates a size distribution of the structure forming the material.

[0033] Specifically, the distribution generating section 104 obtains a Fourier transform result representing a power spectrum by applying Fourier transform to each of the plurality of contrast images generated by the image generating section 102. Then, based on the Fourier transform result corresponding to each of the plurality of contrast images, the distribution generating section 104 generates a size distribution of structures forming the material using a previously known expectation maximization (EM) algorithm.

[0034] Figure 4 is a graph used to describe the Fourier transform results of the contrast image and the estimation of the size distribution. Figure 4 As shown, the distribution generation section 104 applies a previously known Fourier transform to a comparison image Im obtained from a material structure image, and obtains a two-dimensional Fourier transform result P1.

[0035] The horizontal axis and vertical axis of the two-dimensional Fourier transform result P1 represent the frequency. The pixel value in the two-dimensional Fourier transform result P1 represents the intensity of the frequency. Figure 4 The graph P2 shown is Figure 4 The distribution of pixel values ​​of the two-dimensional Fourier transform result P1 in the cross section is shown. The vertical axis of the graph P2 represents the pixel value corresponding to the intensity of the frequency, and the horizontal axis represents the frequency. The points marked as "original" in the graph P2 represent the actual distribution of the pixel values, and the graph marked as "fitted" is a curve that fits the pixel value distribution by a previously known technique.

[0036] The size distribution P3 is generated from the graph P2. Figure 4 In the size distribution P3 shown, "r_dist" represents the size distribution of the structures.

[0037] More specifically, the distribution generating unit 104 generates a size distribution of the structure of the forming material based on the Fourier transform result of the contrast image using the technology disclosed in Japanese Patent Application No. 2019-085277.

[0038] This will be described in detail below.

[0039] Figure 5is a diagram schematically showing a method for estimating the size distribution of structures appearing in a contrast image. Figure 5 As shown, the size of the structure appearing in the contrast image is selected as n The corresponding size s n If the size s n With frequency f i The corresponding probability is P(f i |s n ), then in the following expression, the probability P(f i |S n ) is expressed by I(s,f).

[0040]

[0041] In Bayesian statistics, P(f i |S n ) can be considered as s n is the posterior probability of the prerequisite. Given this assumption, the frequency f i The probability of occurrence P(f i ) can be calculated as the probability P(f i |S n )

[0042] Figure 6 is used to describe the frequency f i The probability of occurrence P(f i ). Figure 6 As shown, the frequency f i The probability of occurrence P(f i ) is related to the frequency f i The corresponding size s m The probability P(f i |s m ), which is summed up by its selection probability π m To weight. The probability P(f i ) is expressed by the following expression.

[0043]

[0044] Figure 7 is a diagram used to describe a method for finding the probability of selection from a frequency distribution. Figure 7 As shown, with and size s n The frequency f in the corresponding Fourier transform result i The probability of the corresponding structure is denoted as P(s n |f i ). Using the previously known Bayesian theorem, the probability P(s n |f i) is expressed by the following expression.

[0045]

[0046] The symbol P(s) in the above expression (3) n ) and the selection probability π n Corresponding frequency f i The probability of occurrence P(f i ) is calculated by the above expression (2). Therefore, with the frequency f i The corresponding structure has size s n The probability P(S n |f i ) is expressed by the following expression, where Expression (2) is substituted into the denominator on the right side of Expression (3), and P(s) in the numerator on the right side of Expression (3) n ) is selected with probability π n Substitute. The symbol z in the following expression (4) in The probability P(s) in expression (3) n |f i )correspond.

[0047]

[0048] Choose probability π n It is expressed by the following expression. In the following expression (5), c i and c j is a predetermined constant.

[0049]

[0050] In step E of the EM algorithm, the distribution generation unit 104 calculates z in expression (4): in In the M step of the EM algorithm, the distribution generation unit 104 calculates the selection probability π in expression (5) n .

[0051] The distribution generation unit 104 repeats the E step and the M step of the EM algorithm until a predetermined convergence condition is satisfied. When the repeated calculation converges, the distribution generation unit 104 generates a size s n The probability of choosing n , to serve as the size distribution of the structures appearing in the contrast image. We choose the probability π n With multiple sizes s n The probability P(s) of the degree of existence (n = 1 to N) n ). Therefore, the probability of choosing n With multiple size S n The size distribution corresponds to (n=1 to N).

[0052] The display section (not shown in the figure) outputs as a result the size distribution generated by the distribution generating section 104. The user of the information processing device 10 views the result of the size distribution displayed at the display section (not shown in the figure).

[0053] The information processing device 10 may be configured to: Figure 8 The computer 50 for realizing the information processing apparatus 10 is implemented by the computer 50 shown in the figure. The computer 50 for realizing the information processing apparatus 10 is provided with a CPU 51, a memory 52 serving as a temporary storage area, and a non-volatile memory unit 53. The computer is further provided with an input / output interface 54 connected to an input / output device or the like (not shown in the figure), and is provided with a read / write unit 55 for controlling the reading and writing of data at a recording medium 59. The computer is also provided with a network interface 56 connected to a network such as the Internet. The CPU 51, the memory 52, the memory unit 53, the input / output interface 54, the read / write unit 55, and the network interface 56 are connected to each other via a bus 57.

[0054] The memory section 53 may be implemented by a hard disk drive (HDD), a solid state drive (SSD), a flash memory, etc. A program that causes the computer to operate is stored in the memory section 53 serving as a storage medium. The CPU 51 reads the program from the memory section 53, loads the program into the memory 52, and sequentially executes the processes included in the program.

[0055] Now, the operation of information processing apparatus 10 according to the first exemplary embodiment is described.

[0056] When the material structure image is input into the information processing device 10, the information processing device 10 executes Fig. 9 The information processing routine shown.

[0057] In step S100 , the acquisition unit 100 acquires an input material structure image.

[0058] In step S102 , the image generating section 102 generates a plurality of comparative images based on the material structure image acquired by the acquiring section 100 , wherein the material structure image is colored according to different contrasts.

[0059] In step S104 , the distribution generation section 104 obtains a Fourier transform result representing a power spectrum by applying Fourier transform to each of the plurality of comparison images generated by the image generation section 102 .

[0060] In step S106 , based on the corresponding Fourier transform results corresponding to the plurality of contrast images, the distribution generating unit 104 uses a previously known EM algorithm to generate a size distribution of the structures forming the material.

[0061] In step S108 , the distribution generating section 104 outputs the size distribution obtained in step S106 as a result.

[0062] As described above, the information processing device 10 according to the first exemplary embodiment acquires a material construction image obtained by imaging a material. Based on the material construction image, the information processing device 10 generates a contrast image, which is a plurality of material construction images having different contrasts. Then, for each of the plurality of contrast images, the information processing device 10 estimates the size of the structure forming the material appearing in the contrast image, and based on the estimated size of the structure, generates a size distribution of the structure forming the material. Therefore, the size distribution of the structure forming the material can be effectively obtained from the construction image of the material. Because the contrast image is generated from the material construction image, the size distribution of the structure appearing only in the contrast image can be obtained.

[0063] Second Exemplary Embodiment

[0064] Now, a second exemplary embodiment is described. The same reference symbols are assigned to the same parts of the image processing apparatus according to the second exemplary embodiment as those of the first exemplary embodiment, and are not described here. The information processing apparatus according to the second exemplary embodiment does not generate a plurality of contrast images corresponding to the material configuration image, but generates a size distribution of structures appearing in this image from a grayscale material configuration image.

[0065] Fig.10 2 is a block diagram showing an example of the functional structure of the information processing apparatus 210 according to the second exemplary embodiment. Fig.10 As shown, the information processing device 210 includes an acquisition unit 100 and a distribution generation unit 204 .

[0066] Without generating a plurality of contrast images with different contrasts, the size distribution of structures appearing in the grayscale material configuration image can be generated from the grayscale image. Therefore, the information processing apparatus 210 according to the second exemplary embodiment does not generate a plurality of contrast images corresponding to the material configuration image, but generates the size distribution of structures appearing in the grayscale material contrast image from the image.

[0067] The distribution generating section 204 generates a Fourier transform result representing a power spectrum by applying Fourier transform to the material configuration image acquired by the acquiring section 100. Then, similarly to the first exemplary embodiment, based on the Fourier transform result of the material configuration image, the distribution generating section 204 employs the EM algorithm to generate a size distribution of structures forming the material.

[0068] Now, the operation of information processing apparatus 210 according to the second exemplary embodiment is described.

[0069] When the material structure image is input into the information processing device 210, the information processing device 210 executes Fig.11 The information processing routine shown.

[0070] In step S100 , the acquisition unit 100 acquires an input material structure image.

[0071] In step S202 , the distribution generation section 204 generates a Fourier transform result representing a power spectrum by applying Fourier transform to the material configuration image acquired in step S100 .

[0072] In step S206 , based on the Fourier transform result obtained in step S202 , the distribution generation unit 204 uses a previously known EM algorithm to generate a size distribution of the structures forming the material.

[0073] In step S108 , the distribution generating section 204 outputs the size distribution obtained in step S206 as a result.

[0074] The other structures and operations of the information processing apparatus according to the second exemplary embodiment are the same as those of the first exemplary embodiment and thus are not described here.

[0075] As described above, the information processing apparatus according to the second exemplary embodiment generates a Fourier transform result representing a power spectrum by applying Fourier transform to a material configuration image, and based on the Fourier transform result, uses the EM algorithm to generate a size distribution of structures forming the material. Therefore, the size distribution of structures forming the material can be effectively obtained from the configuration image of the material. In addition, without generating a plurality of contrast images with different contrasts, the size distribution of structures appearing in the grayscale material configuration image can be generated from this image.

[0076] The processing performed by the apparatus according to the above exemplary embodiments is described as software processing implemented by the program being executed, but the processing can be implemented in hardware. Alternatively, the processing can be combined with both software and hardware. Further, the program stored in the memory section can be stored in any of a variety of memory media and distributed.

[0077] The present disclosure is not limited to the above detailed description. In addition to the above detailed description, it will be apparent that many modifications can be embodied within the technical scope without departing from the gist of the present disclosure.

[0078] For example, in the above-described exemplary embodiments, grains, needles, fibers, holes, etc., of the forming material are mentioned as examples of the structure of the forming material, but different structures are possible.

[0079] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

Claims

1. An information processing device, comprising: Memory; as well as A processor connected to the memory, the processor being configured to: acquiring a material texture image representing the texture of a material, the material texture image being obtained by imaging the material; generating a plurality of comparative images including a plurality of material structure images having different contrasts based on the acquired material structure image; generating a Fourier transform result representing a power spectrum by applying a Fourier transform to each of the plurality of contrast images; as well as Based on the Fourier transform results of the plurality of contrast images, an expectation-maximization algorithm is employed to generate a size distribution of structures forming the material.

2. An information processing method, wherein: The processor performs a process, the process comprising: acquiring a material texture image representing the texture of a material, the material texture image being obtained by imaging the material; generating a plurality of comparative images including a plurality of material structure images having different contrasts based on the acquired material structure image; generating a Fourier transform result representing a power spectrum by applying a Fourier transform to each of the plurality of contrast images; and Based on the Fourier transform results of the plurality of contrast images, an expectation-maximization algorithm is employed to generate a size distribution of structures forming the material.

3. A non-transitory computer-readable storage medium storing an information processing program executable by a processor to perform a process, the process comprising: acquiring a material texture image representing the texture of a material, the material texture image being obtained by imaging the material; generating a plurality of comparative images including a plurality of material structure images having different contrasts based on the acquired material structure image; generating a Fourier transform result representing a power spectrum by applying a Fourier transform to each of the plurality of contrast images; as well as Based on the Fourier transform results of the plurality of contrast images, an expectation-maximization algorithm is employed to generate a size distribution of structures forming the material.

Images